Nano-Watt Modular Integrated Circuits for Wireless Neural Interface.

In this work, a nano-watt modular neural interface circuit is proposed for ECoG neuroprosthetics. The main purposes of this work are threefold: (1) optimizing the power-performance of the neural interface circuits based on ECoG signal characteristics, (2) equipping a stimulation capability, and (3) providing a modular system solution to expand functionality. To achieve these aims, the proposed system introduces the following contributions/innovations: (1) power-noise optimization based on the ECoG signal driven analysis, (2) extreme low-power analog front-ends, (3) Manchester clock-edge modulation clock data recovery, (4) power-efficient data compression, (5) integrated stimulator with fully programmable waveform, (6) wireless signal transmission through skin, and (7) modular expandable design. Towards these challenges and contributions, three different ECoG neural interface systems, ENI-1, ENI-16, and ENI-32, have been designed, fabricated, and tested. The first ENI system(ENI-1) is a one-channel analog front-end and fabricated in a 0.25µm CMOS process with chopper stabilized pseudo open-loop preamplifier and area-efficient SAR ADC. The measured channel power, noise and area are 1.68µW at 2.5V power-supply, 1.69µVrms (NEF=2.43), and 0.0694mm^2, respectively. The fabricated IC is packaged with customized miniaturized package. In-vivo human EEG is successfully measured with the fabricated ENI-1-IC. To demonstrate a system expandability and wireless link, ENI-16 IC is fabricated in 0.25µm CMOS process and has sixteen channels with a push-pull preamplifier, asynchronous SAR ADC, and intra-skin communication(ISCOM) which is a new way of transmitting the signal through skin. The measured channel power, noise and area are 780nW, 4.26µVrms (NEF=5.2), and 2.88mm^2, respectively. With the fabricated ENI-16-IC, in-vivo epidural ECoG from monkey is successfully measured. As a closed-loop system, ENI-32 focuses on optimizing the power performance based on a bio-signal property and integrating stimulator. ENI-32 is fabricated in 0.18µm CMOS process and has thirty-two recording channels and four stimulation channels with a cyclic preamplifier, data compression, asymmetric wireless transceiver (Tx/Rx). The measured channel power, noise and area are 140nW (680nW including ISCOM), 3.26µVrms (NEF=1.6), and 5.76mm^2, respectively. The ENI-32 achieves an order of magnitude power reduction while maintaining the system performance. The proposed nano-watt ENI-32 can be the first practical wireless closed-loop solution with a practically miniaturized implantable device.PhDElectrical EngineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttp://deepblue.lib.umich.edu/bitstream/2027.42/98064/1/schang_1.pd